Effects of environmental enrichment for mice: variation in experimental results.

Abstract

This study focused on the effects of different enriched environments for mice in a number of behavioral and physiological parameters in 2 routine laboratory testing procedures: potency testing for tetanus vaccine and stress-induced hyperthermia. The variability in the results was studied by calculating and analyzing mean absolute devi-ations. Mice from enriched conditions weighed more and consumed more food than mice from standard housing conditions. However, mice from enriched conditions lost more body weight after being housed individually. Other physiological parameters showed no differences. Mice from standard conditions were more active in an open field, suggesting a tendency to overrespond to various stimuli in a testing environ-ment. Mice from enriched environments were more tranquil and easier to handle. The enrichment did not influence the variability in any of the parameters measured, al-though earlier results and results of other studies suggest that the effects on the vari-ability in results are parameter dependent. When enrichment does not influence vari-ability, there is no reason for not introducing cage enrichment and by doing so contributing to the animals' welfare.

Full text

Page 1

Effects of Environmental Enrichment
for Mice: Variation
in Experimental Results
Heleen A. Van de Weerd and Emma L. Aarsen
Department of Laboratory Animal Science
Utrecht University
Anne Mulder
Department of Laboratory Animal Science
Solvay Pharmaceuticals
Weesp, The Netherlands
Cas L. J. J. Kruitwagen
Center for Biostatistics
Utrecht University
Coenraad F. M. Hendriksen
National Institute of Public Health and the Environment
Bilthoven, The Netherlands
Vera Baumans
Department of Laboratory Animal Science
Utrecht University
This study focused on the effects of different enriched environments for mice in a
number of behavioral and physiological parameters in 2 routine laboratory testing
procedures:potencytestingfortetanusvaccineandstress-inducedhyperthermia.The
variabilityintheresultswasstudiedbycalculatingandanalyzingmeanabsolutedevi-
JOURNAL OF APPLIED ANIMAL WELFARE SCIENCE, 5(2), 87–109
Copyright © 2002, Lawrence Erlbaum Associates, Inc.
Requests for reprints should be sent to Heleen A. Van de Weerd, ADAS Terrington, Terrington St.
Clement, Norfolk, England PE34 4PW. E-mail: heleen.vandeweerd@adas.co.uk

Page 2

ations. Mice from enriched conditions weighed more and consumed more food than
micefromstandardhousingconditions.However,micefromenrichedconditionslost
more body weight after being housed individually. Other physiological parameters
showed no differences. Mice from standard conditions were more active in an open
field, suggesting a tendency to overrespond to various stimuli in a testing environ-
ment.Micefromenrichedenvironmentsweremoretranquilandeasiertohandle.The
enrichment did not influence the variability in any of the parameters measured, al-
though earlier results and results of other studies suggest that the effects on the vari-
ability in results are parameter dependent. When enrichment does not influence vari-
ability, there is no reason for not introducing cage enrichment and by doing so
contributing to the animals’ welfare.
Environmental enrichment, or modifications to the environments of animals to
improve their biological functioning (Newberry, 1995), increasingly has been
introduced into laboratory animal research facilities. From a welfare point of
view, this is a good development, because providing environmental enrichment
improves the animal’s well-being. Enriched environments release and structure
species-specific behavior, meeting more of the behavior needs (Beaver, 1989;
Benn, 1995; Chamove, 1989; Fortmeyer, 1982; Newberry, 1995; Van de Weerd,
1996). Especially when the design of the enriched environment takes into ac-
count specific behavioral demands of the animals as assessed by means of pref-
erence tests (Mench, 1994; Van de Weerd, Van Loo, Van Zutphen, Koolhaas, &
Baumans, 1997b), the improved housing conditions may enhance the welfare of
the animals. When deprived of the possibility of performing species-specific be-
havior, animals may show signs of suffering, such as abnormal behavior or other
pathology (Jensen & Toates, 1993). Providing environmental enrichment, such
as for building nests, gives the animals more control over their environment by
structuring it (Beaver, 1989; Benn, 1995; Chamove, 1989; Van de Weerd, Van
Loo, Van Zutphen, Koolhaas, & Baumans, 1997b; Van de Weerd, Van Loo, Van
Zutphen, Koolhaas, & Baumans, 1998a, 1998b). Increasing the controllability of
relevant events may reduce any stress experienced by an animal (Wiepkema &
Koolhaas, 1993).
Thereseemstobesomeconcern,however,astowhetherenvironmentalenrich-
ment conflicts with the standardization of experiments. Standardization increases
the reproducibility and comparability of experiments. It aims at reducing un-
wantedvariationcausedbyanimalandenvironmentalfactorsandthusreducesthe
number of animals needed in experiments (Beynen, Gärtner, & Van Zutphen,
1993;Gärtner,1999;VanZutphen,Hedrich,&VanOortmerssen,1993).Somere-
searchers fear that “enriched” animals show more variability in their responses to
experimental procedures because they show more diverse behavior. In complex
environments, animals are responding not just to one stimulus in isolation but to
many variable stimuli at once (Appleby, 1997), which may cause increased varia-
88
VAN DE WEERD ET AL.

Page 3

tion within subjects (Eskola, Lauhikari, Voipio, Laitinen, & Nevalainen, 1999) or
enhance the deviation in experimental data (Gärtner, 1999).
One might argue that because animals can perform more of their species-spe-
cific behavior in enriched environments, they may be better able to cope with
novelandunexpectedchanges,thusshowingauniformresponse(Baumans,1997;
Chamove, 1989; Rose, 1994; Van de Weerd, 1996; Wemelsfelder, 1994). More-
over, animals living in an enriched environment may be less excitable than ani-
mals in barren environments because restricting sensory input makes the nervous
system more sensitive and more reactive to external stimulation (Grandin, 1989).
Also, restrictive rearing appears to produce deficits in learning ability as well as
tendencies to overrespond to stimuli in a new environment (Joseph & Gallagher,
1980). Consequently, animals from enriched housing conditions would be ex-
pected to be more stable physiologically and psychologically and may, therefore,
be considered more refined animal models, ensuring better scientific results
(Bayne, 1996; Benn, 1995; Rose, 1994; Spinelli & Markowitz, 1985; Van de
Weerd, 1996). They might be more suitable as models for humans, as it is ques-
tionable whether keeping animals in nonstimulating, standard environments
makesthemadequatemodels(Markowitz&Gavazzi,1995).Ifhousingconditions
do not meet the demands of a particular species, one cannot expect reliable and re-
producible results (Fortmeyer, 1982).
If one provides animals in the laboratory with the very best physical and social
environmental conditions for their well-being, then one should need to use fewer
of them in research experiments or routine tests, and one’s results will be accurate
andreliable.Intheseoptimalconditions,variationwillbegreatlyreduced(Chance
& Russell, 1997; Russell, 1994). This is not a new observation: In the 1950s,
Chance(1956,1957)demonstratedthatthesizeofthevarianceisrelatedtothena-
tureoftheconditions—housing,treatment,andsocialsituation—inwhichtheani-
mal in the laboratory is kept. As many researchers have reported (e.g., Van de
Weerd, 1996), this means that enrichment influences not only group means of
measured parameters but also the variability of these results. If these expectations
are true across a broad range of experiments, then eventually the use of enriched
animals in the laboratory may lead to a reduction in the number of animals neces-
sary for obtaining valid experimental results. However, few studies have focused
specifically on the variability in results. Moreover, studies use different methods
of analysis, such as coefficients of variation per parameter (Gärtner, 1999) or of
different groups (Tsai & Hackbarth, 1999) or Solo power analysis (Eskola et al.,
1999).Inthisstudywecalculatedandanalyzedmeanabsolutedeviations(MADs)
to compare the variability of individual mice, instead of groups of mice, from dif-
ferent environments (meaning a lower N). We calculated the MAD as the (posi-
tive) difference between the individual value for one mouse and the mean value
from that mouse’s group or cage. Zimmerman (1999) used a comparable method
for rats, using medians instead of means.
EFFECTS OF ENVIRONMENTAL ENRICHMENT FOR MICE
89

Page 4

In this study we investigated the effects of enrichment on the results of a num-
ber of behavioral and physiological parameters in two routine testing procedures.
The selection of the cage enrichments we used (nesting material and nestbox) was
based on preference tests conducted by Van de Weerd and colleagues (Van de
Weerd et al., 1997b, 1998a, 1998b) and on practical use and previous experience
(wire grid floor, plastic tube, gnawing blocks). The first experiment, in which we
studied variability in antiserum responses of mice used in potency testing of teta-
nusvaccine,wascarriedoutattheNationalInstituteofPublicHealthandtheEnvi-
ronment(RIVM;Bilthoven,TheNetherlands).Apilotexperiment(VandeWeerd,
Willems,Hendriksen,&Baumans,1999)inwhichthesameprocedureswereused
showednosignificanteffectsofthehousingconditionsontheimmuneresponseor
the variation in the immune response. In this study, which is a follow-up of the pi-
lot experiment, we used three different housing conditions: tissue, enriched, and
superenriched. The provision of a paper tissue is the routine housing condition at
the RIVM and, therefore, was included in this study. Previous research into the ef-
fects of tissues as enrichment showed no major effects of enrichment on behav-
ioral and physiological parameters (Van de Weerd et al., 1997a). Because the
degree and form of the enrichment might be important for the size of the effects
(VandeWeerd,1996),weincludedaconditionwithahigherdegreeofenrichment
(superenriched) to act as a positive control. We investigated two different rearing
conditions, standard and tissue, to see if early rearing conditions affected immune
responseorbehavior.Wealsotestedmiceinanopen-fieldtesttoobtainadditional
information about their behavior in a novel environment. The second experiment,
in which we tested mice from standard and superenriched conditions, was con-
ducted at a pharmaceutical company, Solvay Pharmaceuticals BV, (Weesp, The
Netherlands). We investigated the variability in the results of a physiological rou-
tine procedure: stress-induced hyperthermia (SIH), which is normally used to de-
tectpossibleanxiolyticeffectsofdrugsinmice.Effectsofenrichmentonhandling
of the animals also were studied in this experiment. Because housing conditions
may have an effect on body weight and food intake (Chvédoff, Clarke, Irisarri,
Faccini, & Monro, 1980; Van de Weerd et al., 1997a; Van de Weerd et al., 1999),
we monitored these in both studies.
ANIMALS, MATERIAL, AND METHOD
Experiment 1
Animals.
Bilthoven,TheNetherlands).Atthestart,theanimalswereapproximately3weeks
of age and weighed 10 to 14 g. Half of these mice (n = 64) had been reared under
“tissue” conditions (bedding and one tissue, Kleenex©, Kimberly-Clark Corpora-
The sample comprised 128 male mice (RIVM:N:NIH; RIVM,
90
VAN DE WEERD ET AL.

Page 5

tion,Veenendaal,TheNetherlands),andtheotherhalf(n=64)hadbeenrearedun-
der standard conditions (only sawdust bedding). After arrival, the mice were
housedingroupsof8.Oneanimalfromtheenriched-housingconditiondiedduring
the fifth week of the experiment.
Housing conditions.
enriched,andsuperenriched.Thetissuecageswerewire-toppedMakrolonTypeII
cages(375cm2;UNOroestvaststaal,Zevenaar,TheNetherlands)witha1-cmlayer
ofsawdustbedding(10/20BK;BMI,Helmond,TheNetherlands)andoneKleenex
tissue. The enriched cages were wire-topped Makrolon Type II cages with a 1-cm
layer of sawdust bedding, one Kleenex tissue, and a nest box of perforated sheet
metal(8cm×10cm×6cm)withametalclimbinggrid(8cm×9cm)attachedtoit.
Thesuperenrichedmicewerehousedinwire-toppedMakrolonTypeIIIcages(840
cm2;UNOroestvaststaal,Zevenaar,TheNetherlands)witha1-cmlayerofsawdust
bedding, two Kleenex tissues, the nest box with climbing grid, two Aspen wood
gnawingblocks(large:2cm×2cm×10cm,small:1cm×1cm×5cm;TapveiOy,
Kaavi, Finland), a plastic tube (PVC; Ø 7 cm, length: 16 cm), wood–wool (±5
gramsTapveiOy,Kaavi,Finland),andastainlesssteelwiregridfloor(16cm×11
cm) under the food hopper.
In all groups, food pellets (2122 SRM–A 10µm 0.9M RAD, Hope Farms,
Woerden, The Netherlands) and tap water were provided ad libitum. The room
temperature was 20 to 24°C, relative humidity was approximately 60%, light in-
tensity varied between 210 lux (at floor level) and 240 lux (at the top cage level).
The lights were on from 7:00 a.m. until 7:00 p.m. The ventilation rate was 7 air
changes per hour, and a radio played softly in the background.
Three different housing conditions were used: tissue,
Experimental design.
housed (after weaning) in groups of 8 on the basis of their body weight, so that the
mean body weight of the groups was equal. Within a rearing condition, the cages
weredividedrandomlyoverthethreehousingconditions.Twogroupsof8animals
were housed under the tissue conditions, three groups were housed under enriched
conditions, and three groups were housed under superenriched conditions. Mice
were individually marked. Twice a week, the cages were cleaned and tissues and
wood–wool renewed and group food intake was measured by weighing the food.
On the second day of the experiment the mice were injected with tetanus toxoid.
Thetotaldurationoftheexperimentwas5weeks.Threedaysbeforetheendofthe
experiment, the animals were subjected to an open-field test.
On the first day of the experiment, the mice were
Tetanus vaccine potency test.
activatedvaccineproducedisthedemonstrationofpotencyoftheseproductsinan
animal model. The potency test of the tetanus toxoid vaccine is based on a
serologicalmethod.Groupsofmiceareimmunizedwithserialdilutionsofthevac-
Part of the quality control of batches of in-
EFFECTS OF ENVIRONMENTAL ENRICHMENT FOR MICE
91

Page 6

cine under study and a reference preparation, respectively. After 5 weeks, animals
arebled,andserumsamplestitratedintheinvitroToxinBindingInhibition(ToBI)
test.InTheNetherlands,thistesthasbeenacceptedofficiallybytheregulatoryau-
thorities and has been used routinely since January 1996.
Inthisexperimentonlyonedilution(12.5µm/ml)ofthetetanustoxoidprepara-
tion (0.5 ml of DTA 93/1) was used. At the end of the experimental period, mice
were killed by terminal bleeding under halothane anesthesia and serum prepared.
The serum samples were analyzed in the Tetanus ToBI test (Hendriksen, Van der
Gun, Nagel, & Kreeftenberg, 1988), and this test was used to estimate the immune
response.
Open-field test.
andthemiceweredividedintotwoequalgroups.Themicefromthesetwogroups
were divided evenly and allocated to morning and afternoon groups to rule out
possible diurnal effects. The test was carried out in the animal room, and a video
system was used to record each test; thus, the experimenter did not have to be
present in the testing room. Afterward, the behavior of the mice was scored from
the videotape using a behavioral observation software package (Observer®;
1990). The ethogram used (see Table 1) was based on the pilot study (Van de
Weerd et al., 1999). The open-field arena consisted of a grey PVC wall (50 cm
high)surroundingthecircularopenfield(Ø90cm).TwoclearPerspexV-shaped
objects with holes (Cheese Slice, IMS, Cheshire, England) were placed upside
down 25 cm from the center and opposite each other. The illumination level var-
ied between 80 and 170 lux at ground level. An individual mouse was placed in
the center of the open field between the two objects. Behavior was recorded for 5
min. Between two tests the apparatus and the objects were cleaned with ethyl al-
cohol (90%) soaked tissues.
The open-field test was performed on 2 consecutive days,
92
VAN DE WEERD ET AL.
TABLE 1
Experiment 1: Ethogram of the Open-Field Test
BehaviorDefinition
Locomotion
Immobility
Walking with four feet on the floor
No body movements for one s or longer, head movements are allowed
(includes also freezing: no movement at all)
Sniffing or gnawing at the object, rearing against it, sitting or walking
under or through it
Climbing on the object (with four feet off the ground)
Standing on the hind legs or leaning against the wall
Grooming (licking, scratching), sniffing, walking only with the front
legs (hind part of the mouse on the same spot)
Interaction with the object
Climbing
Rearing
Other

Page 7

Experiment 2
Animals.
(BALB/cANnCrl BR, Charles River Wiga, Sulzfeld, Germany). At the start of the
experiment they were 3 weeks of age. They were reared without any enrichment.
Afterarrival,theyacclimatizedintheanimalroomfor2daysandthenwerehoused
in groups of 5. One animal from the enriched-housing conditions died during the
fifth week of the experiment.
The Experiment2 samplecomprised 50malemice
Housing conditions.
superenriched conditions. Standard housing conditions were wire-topped
Makrolon Type III cages (840 cm2; Tecniplast, Rome, Italy) with a 2-cm layer of
sawdust bedding (Woody Clean S 8/15, Rettenmaier, Germany). Superenriched
cageswereMakrolonTypeIIIcages(840cm2;Tecniplast,Rome,Italy)withsaw-
dustbedding,twoKleenextissues(Kimberly-ClarkCorporation,Veenendaal,The
Netherlands), a nest box of perforated sheet metal (8 cm × 10 cm × 6 cm) with a
metal climbing grid (8 cm × 9 cm), two Aspen wood gnawing blocks (Tapvei Oy,
Kaavi, Finland, large: 2 cm × 2 cm × 10 cm, small: 1 cm × 1 cm × 5 cm), a plastic
tube(PVC,Ø7cm,length:16cm),papernestingmaterial(Enviro-dri±10g;BMI,
Helmond, The Netherlands), and a stainless steel wire grid floor (16 cm × 11 cm,
meshsize10mm2×10mm2)underthefoodhopper.Foodpellets(RMH–TM;Hope
Farms,Woerden,TheNetherlands)andwaterwereprovidedadlibitum.Theroom
temperature was 20 to 21°C, and the relative humidity was approximately 60%.
The lights were on from 7:00 a.m. until 7:00 p.m. Ventilation rate was 16 air
changes per hour, and a radio played softly in the background.
Two housing conditions were used: standard and
Experimental design.
housed in groups of 5 on the basis of their body weight, so that the mean body
weight of the groups was equal. The cages were divided randomly over the two
housing conditions. Mice were marked individually with a waterproof text marker
on the tail, and this was renewed every week after the mice were weighed. Every
week,thecageswerecleaned,tissuesandpapernestingmaterialwererenewed,and
group food intake was measured by weighing the remaining food. The total dura-
tion of the experiment was 6 weeks; after 5 weeks of housing under the different
conditions the mice were tested in the SIH test. For this procedure, the mice had to
be housed individually for approximately 24 hr. We measured body weight before
andafterthisisolationtoseeifmicefromdifferenthousingconditionsreacteddif-
ferently to isolation.
On the first day of the experiment, the mice were
SIH.
Measurements of (rectal) body temperature at repeated intervals cause
the body temperature of mice to rise, thus indicating stress. The procedure, de-
scribed by Van der Heyden, Zethof, and Olivier (1997), can test the effects of dif-
EFFECTS OF ENVIRONMENTAL ENRICHMENT FOR MICE
93

Page 8

ferent pharmaceutical compounds as well as the reactions of mice to different cir-
cumstances, such as housing conditions. Rectal temperatures were measured at 0
min(basaltemperature)and10minlater(theincreaseintemperatureisastressre-
sponsetothehandlingofthemouse).Themeasurementwascarriedoutbyholding
a mouse in one hand (45° angle); with the other hand, a thermistor probe was in-
serted gently for a length of 2 cm into the rectum of a mouse. Digital recordings of
thetemperatureweredeterminedwithanaccuracyof0.1°CbymeansofaKeithley
871Adigitalthermometer(NiCr–NiAlthermocouple).Theprobe,dippedintosili-
con oil before inserting, was held into the rectum until a stable rectal temperature
wasmeasuredfor20sec.Temperatureintheexperimentalroomwas20°C.Testing
was performed between 10:00 a.m. and 12:00 p.m. and 1:00 p.m. and 3:00 p.m.
Groups of different housing conditions were evenly allocated to morning or after-
noon sessions.
Body weight after isolation.
individually in Makrolon Type I cages (216 cm2; UNO Roestvaststaal, Zevenaar)
forapproximately24hraftertransportationtotheroomwheretheSIHtestwasgo-
ingtobeperformed.Theywereweighedonadigitalbalancebeforeandafterisola-
tion for the SIH procedure.
For the SIH procedure, the mice were housed
Handling.
ing weighing (a mouse was placed in a small cage on the weighing scale), and tail
marking(mousewasheldbythetailandtextmarkerwasapplied)toevaluatetheir
behavior during these procedures. Five phases during these procedures were ob-
servedandclassifiedintotwoorthreecategories(seeTable2).Thescoregivenwas
anagreementscoregivenbytwoindependentobserverswhowerenotfamiliarwith
the housing conditions of the mice (not visible on the videotape).
In the last week of the experiment, the mice were videotaped dur-
Statistical Analyses for Experiments 1 and 2
We analyzed all collected data for differences in both mean and MAD (varia-
tion), using the statistical package SPSS (Version 9.0). Data of individual mice
were analyzed; however, “group” was included in the analysis as a factor, taking
into account possible variation between the groups. To analyze possible differ-
ences in MAD, we subtracted the mean value per group or cage from the indi-
vidual value of a mouse: MAD =−
XX
i
ses, similar to those for the means, on the absolute values of this subtraction.
Wesubjectedindividualbodyweights,bodyweightsbeforeandafterisolation,
foodintakepercage,andSIHtoarepeatedmeasuresanalysis(multivariatePillai’s
test) to detect time effects and possible differences between the housing condi-
tions.Weusedamultivariateanalysisofvariance(MANOVA)toanalyzepossible
. We then carried out statistical analy-
94
VAN DE WEERD ET AL.

Page 9

effects of the housing conditions on the duration and frequencies of the behavioral
elements from the open-field test. To determine which of the four behavioral ele-
ments (climbing was omitted because of its low frequency) differed between the
housing conditions, we subsequently carried out a Bonferroni-corrected analysis
of variance (ANOVA) on each of them. To compare the outcome of the ToBI test,
we chose an arbitrary optical density (OD) level of 0.700, just above the OD50
value(theOD50valueisanindirectmeasureforthatdilutionoftheseruminwhich
halfoftheaddedtetanustoxinhasbeenneutralizedbytheantibodiespresentinthe
serum). For each mouse, we calculated the extent to which the serum had to be di-
lutedtoachieveanexactopticaldensityof0.700.Wesubjectedthesetheoretically
calculated values to a MANOVA.
Whenwefoundoveralldifferencesamongthethreehousingconditions(RIVM
experiment), we made multiple comparisons (contrasts) using the Bonferroni cor-
rection to determine which differences were significant. Nonparametric scores for
the handling procedures were analyzed with a Mann–Whitney U test. The overall
level of statistical significance was preset at p < .05.
RESULTS
Experiment 1
Body weight and variation in body weight.
weight of the mice in each housing condition and their MADs. Repeated measures
Figure 1 shows mean body
EFFECTS OF ENVIRONMENTAL ENRICHMENT FOR MICE
95
TABLE 2
Experiment 2: Behavior Scored During Weighing and Tail Marking Procedures
Phase Scores
Behavior in weighing cage 1 = Mouse shows few movements and has normal breathing, is
explorative
2 = Mouse shows more movements and faster breathing, may
show rearing
Behavior during tail marking
First part 1 = Mouse shows few movements, breathing is normal
2 = Mouse “walks” with forepaws
3 = Mouse “walks” with all four limbs, wants to get away
Same scores as for first part
1 = Mouse does not turn around
2 = Mouse turns around and its position needs to be corrected
1 = Very relaxed, at ease, few movements, shows normal
breathing
2 = Stressed, freezing behavior, fast breathing
Second part
Turning of mouse
Overall impression of mouse

Page 10

analysis found no overall significant housing effect but did find a significant time
effect (p < .001). A Time × Housing interaction (p < .001) was found. In the first
weeks of the experiment, tissue housed mice gained weight faster; but from Week
3Bon,micefromsuperenrichedhousingconditionsgainedweightfasterthanmice
fromthetissueandenrichedhousingconditions.(PartAoftheweeksconsistedof3
days;PartBoftheweeksconsistedof4days.)Micefromtheenrichedhousingcon-
ditionalsogainedweightfasterthanthemicefromthetissueconditions.ATime×
Housing group interaction (p < .01) was found, which means that some groups
gainedweightfasterthanothersdid,butthiswasnotspecificforanyofthehousing
conditions.
TheMAD(seeFigure1,bottom)betweenhousingconditionsdidnotdiffer,but
again there was a significant time effect (p < .01). Repeated measures analysis
showed a significant effect for rearing (p < .05, data not shown) and a Time ×
Rearing interaction (p < .001). Mice reared in tissue cages were heavier (from the
beginningoftheexperiment)andgainedweightfasterthanthemicerearedinstan-
dard cages. No differences were found in the MAD, but there was a significant
time effect for rearing (p < .01).
Food intake.
fect (p < .001) for food intake (data not shown). Comparisons revealed that mice
fromsuperenrichedhousingconditionsconsumedsignificantlymorefoodthanen-
riched(p<.001)andtissuehousedmice(p<.001),independentoftherearingcon-
Repeated measures analysis showed a significant housing ef-
96
VAN DE WEERD ET AL.
FIGURE 1
dev.;bottom)perhousingcondition.PartAoftheweeksconsistedof3days,PartBof4days.N
= 127. S-enriched = superenriched.
Experiment 1: Mean body weight (top) and mean absolute deviations (Mean abs.

Page 11

ditions. A significant time effect (p < .01) was found, indicating that overall food
consumption increased with time.
The effect of rearing conditions was significant (p < .01); mice reared with tis-
sues consumed more food than mice reared in standard conditions. No MADs
could be calculated because food intake was measured per group of mice.
Open-field test duration of behavioral elements and variation in
duration.
Figure 2 shows the mean duration of the behavioral elements of the
open-field test for the three different housing conditions and their mean absolute
deviations.TheMANOVAshowedanoverallsignificanthousingeffect(p<.001).
Further analysis (Bonferroni-corrected ANOVAs) revealed that duration of loco-
motionandinteractionwiththeobjectweresignificantlydifferentamongthehous-
ing conditions (locomotion: p < .001, interaction: p < .001). Tissue housed mice
showed more locomotion than the enriched mice (multiple comparisons, p < .001)
and superenriched housed mice (multiple comparisons, p < .001). Enriched mice
alsoshowedsignificantlymorelocomotionthansuperenrichedhousedmice(mul-
tiplecomparisons,p<.01);however,micefromsuperenrichedconditionsshowed
moreinteractionwiththeobjectsthandidthetissuehousedmice(multiplecompar-
isons,p<.001)andtheenrichedhousedmice(multiplecomparisons,p<.001).En-
riched housed mice showed more interaction than tissue housed mice (multiple
comparisons, p < .001).
EFFECTS OF ENVIRONMENTAL ENRICHMENT FOR MICE
97
FIGURE2
deviations(Meanabs.dev.;bottom).N=127.LOCO=locomotion;INT=interaction;CLIM=
climbing; REAR = rearing; OTHER = other behavior; S-enriched = superenriched.
Experiment1:Meandurationofopen-fieldbehavior(top)andtheirmeanabsolute

Page 12

No significant effects were found for the MADs among the three different
housingconditions.Rearingconditionsdidnotshowanysignificantdifferencesin
the duration of the behavioral elements or their MADs.
Open-field test frequency of behavioral elements and variation in
frequency.
Figure 3 shows the mean frequencies of the behavioral elements of
the open-field test for the three different housing conditions and their mean abso-
lute deviations.
TheMANOVAshowedanoverallsignificanthousingeffect(p<.001).Further
analysis (Bonferroni-corrected ANOVAs) revealed that frequency of locomotion
andinteractionwiththeobjectweresignificantlydifferentamongthehousingcon-
ditions (locomotion: p < .01, interaction with object: p < .001). Mice from the
superenriched conditions showed a higher frequency of locomotion (multiple
comparisons, p < .01) and interaction (multiple comparisons, p < .001) than mice
from enriched conditions and a higher frequency of locomotion (multiple compar-
isons,p<.01)andinteraction(multiplecomparisons,p<.001)thanmicefromtis-
sue conditions.
No significant effects were found for the MADs among the three different
housing conditions. Rearing conditions did not cause significant differences in the
frequency of the behavioral elements or their MADs.
98
VAN DE WEERD ET AL.
FIGURE3
viations (Mean abs. dev.; bottom). N = 127. LOCO = locomotion; INT = interaction; CLIM =
climbing; REAR = rearing; OTHER = other behavior; S-enriched = superenriched.
Experiment1:Meanfrequencyofopen-fieldbehavior(top)andmeanabsolutede-

Page 13

Immune response and variation in immune response.
ANOVA revealed no significant differences for housing conditions in immune re-
sponse or their MADs (see Table 3) or for the rearing conditions. A small effect of
housing groups (p < .05) was found: In some groups, variation was smaller than in
other groups, but this was not for a specific housing condition.
The univariate
Experiment 2
Body weight and variation in body weight.
body weight of the mice per housing condition and their MADs. Repeated mea-
sures analysis found an overall significant housing effect (p < .001): Mice from
superenriched housing conditions were heavier than those from standard condi-
tions. A significant time effect was also found (p < .001). A Time × Housing inter-
action (p < .001) was found, probably caused by the fact that mice from the
superenriched housing conditions gained weight faster than mice from standard
housingconditions.TheMADsamonghousingconditionsdidnotdiffer,butthere
was a significant time effect (p < .01).
Figure 4 shows the mean
Food intake.
.01) effect (data not shown). Mice from superenriched housing conditions con-
sumed significantly more food than did standard housed mice. A significant time
effect (p < .001) was also found, indicating that overall food consumption varied
over time.1
Repeatedmeasuresanalysisshowedasignificanthousing(p<
Body weight after isolation and variation in body weight.
showsmeanbodyweightsofthemicebefore(Weight1)andafter(Weight2)being
housedindividuallyforapproximately24hr(isolation)andtheirMADs.Repeated
measuresanalysisfoundanoverallsignificanthousingeffect(p<.001):Micefrom
superenriched housing conditions had overall higher body weight (Weight 1 and
Weight 2 together).
A significant time effect also was found (p < .001). A Time × Housing interac-
tion(p<.05)wasfound,indicatingthatmicefromthesuperenrichedhousingcon-
ditions lost slightly more weight than mice from standard housing conditions.
Because mice from superenriched conditions were significantly heavier than mice
from standard conditions, they had more weight available to lose. Therefore, we
also analyzed relative decrease in body weight and confirmed that mice from
superenriched conditions lost more weight than standard housed mice (p < .05).
The MADS among housing conditions did not differ.
Figure5
EFFECTS OF ENVIRONMENTAL ENRICHMENT FOR MICE
99
1Nomeanabsolutedeviationscouldbecalculatedbecausefoodintakewasmeasuredpergroupofmice.

Page 14

SIH and variation in SIH.
miceatTime0(SIH0)and10minlater(SIH10)andtheirMADs.Nooverallsignif-
icant housing effect or Time × Housing interactions were found with the repeated
measuresanalysis.Asignificanttimeeffectwasfound(p<.001).Themicereacted
to the procedure with a rise in body temperature, which was similar for mice from
both housing conditions. The MADs among housing conditions did not differ, but
there was a significant time effect (p < .001).
Figure 6 shows mean body temperatures of the
Handling.
U tests revealed a significant housing effect in the total handling score, which was
thesumofallscorestogether,exceptfortheoverallimpressionscore(p<.01).The
Table4showstheresultsforthehandlingscores.Mann–Whitney
100
VAN DE WEERD ET AL.
FIGURE 4
dev.; bottom) per housing condition. N = 49. S-enriched = superenriched.
Experiment 2: Mean body weight (top) and mean absolute deviations (Mean abs.
TABLE 3
Experiment 1: Mean Values of Immune Response Test and Their Mean Absolute
Deviations for the Different Housing Conditions
Housing ConditionImmune ResponseMean Absolute Deviation
Tissue
Enriched
Superenriched
.82
.77
.91
.26
.28
.38

Page 15

separatescoresthatreachedsignificancewerethebehaviorintheweighingcage(p
< .01) and the overall impression of the state of the mouse (p < .01). Behavior dur-
ingthefirstphaseoftailmarkingshowedatrend(p=.053).Duringallthesephases,
mice from superenriched housing conditions had lower scores, indicating they
were more at ease than mice from standard housing conditions.
DISCUSSION
The enrichment used in this study influenced several parameters. In both experi-
ments, body weight and food intake were affected. The effect on body weight
was largest in Experiment 2; mice from the superenriched conditions weighed
more than mice from the standard housing conditions. In Experiment 1 the effect
was smaller, but mice from superenriched conditions and enriched conditions
gained weight faster than mice from tissue housing conditions. Other researchers
also have found a significant increase in body weight for enriched housed mice
(Chapillon, Manneché, Belzung, & Caston, 1999; Dahlborn, Van Gils, Van de
Weerd, Van Dijk, & Baumans, 1996; Manosevitz & Joel, 1973; Van de Weerd,
EFFECTS OF ENVIRONMENTAL ENRICHMENT FOR MICE
101
FIGURE 5
(Weight2)beinghousedindividuallyforapproximately22hr(isolation)andmeanabsolutede-
viations (Mean abs. dev.; bottom). N = 49. S-enriched = superenriched.
Experiment 2: Mean body weight (top) of the mice before (Weight1) and after